Combined FTIR and X-ray diffraction for estimating heat-induced metric skeletal changes using chemosteometric regression models

The chemosteometric regression models proposed by Gonçalves et al. (2020) based on data obtained through Fourier-transform infrared spectroscopy in attenuated total reflectance mode provided encouraging results regarding the prediction of heat-induced metric changes in human bone. In this study, the potential of c hemosteometry was further explored by adding powder X-ray diffraction (XRD) data. Bones from six experimentally burnt human skeletons from the 21st Century Identified Skeletal Collection (four females and two males with ages at death between 62 and 88 years) were measured before and after burning. Cortical bone was sampled from each bone and analysed through FTIR-ATR and XRD. The resulting data were used to build regression models to predict the percentage of heat-induced metric changes. Results were then compared to the actual metric changes. Chemometric data that most significantly correlated with such changes were the infrared band at 3572 cm −1 (hydroxyapatite O H stretch) and the XRD crystallite size. The infrared crystallinity index also provided relevant information for the regression models. Adjusted R 2 of the regression models ranged between 0.20 and 0.68 and mean absolute error (MAE) values ranged between 2.59 and 4.22 percentage points. Models based only on the maximum length of clavicles, metacarpals and metatarsals provided even better predictions, suggesting that results vary according to bone type. In contrast, correction factors of 10%, 12% and 15% resulted in higher MAE of 5.50, 6.34 and 8.20 percentage points in the same sample, respectively. The added benefit of adding the XRD crystallite size to models including FTIR data was ambiguous, appearing to be irrelevant for the overall sample, but improving estimations in the subsample that only comprised bones with diaphysis. The results suggest that the prediction of heat-induced shrinkage in epiphyseal measurements is to some extent undermined by using FTIR and XRD data obtained from cortical bone samples. This study concluded that chemosteometry provides an objective case-by-case prediction of heat-induced shrinkage, having clear advantages over the shrinkage correction factor approach.